Tertium datur: Multi-attribute reference points and integration choices between the European Union and Eurasian Economic Union

The article imports insights from prospect theory into the study of integration choices of ruling elites from Eastern Partnership countries. It introduces the notion of multi-attribute reference points and provides an example of identifying their coordinates, against which ruling elites from Armenia, Azerbaijan, Belarus, Georgia, Moldova and Ukraine are expected to consider distinct integration choices: the European Union or Eurasian Economic Union. The research finds that ruling elites from Eastern Partnership countries with the lowest levels of affluence, with medium to high intensity conflicts with Russia and with lower, but still non-trivial costs of domestic transformation have tended to be risk-seeking and opted for the European Union as an integration choice. However, ruling elites from Eastern Partnership countries with low and medium levels of affluence, with no conflict with Russia and with medium to high costs of domestic transformation have tended to be risk-averse and selected the Eurasian Economic Union as an integration option.


Introduction
After fully completing negotiations over the content of the Association Agreement (AA) with the European Union (EU), Armenian President Serzh Sargsyan took many observers by surprise when he announced in September 2013 that Armenia was suspending the association process to the EU and opting instead for integration into the Eurasian Economic Union (EAEU). Ukraine experienced a double U-turn in a period of several months. After negotiating 4 years, from 2007 to 2011, and preparing for two more years for the signature of AA with the EU, Viktor Yanukovych announced in November 2013 the suspension of association process. The new post-Maidan administration under Petro Poroshenko reversed the decision of Yanukovych's entourage and signed the AA with the EU in the first-half of 2014.
Ruling elites from Georgia and Moldova have opted for the EU as an integration project, whereas Belarus is a founding member of the EAEU from the onset. For the time being, Azeri elites have selected the third alternative of staying away from both regional integration projects.
The existing scholarship tends generally to adopt outside-in and inside-out perspectives in explaining the choice between the EU and EAEU as regional integration projects. Mearsheimer (2014), Sakwa (2016) and, to some extent, Charap and Colton (2017) take the outside-in perspective and argue in line with the realist tradition that strategic competition between major external players, such as the United States, EU and Russia, has forced the ruling elites from Eastern Partnership (EaP) countries to make a choice between the EU and EAEU. Wilson (2014), Dragneva and Wolczuk (2015) and Kuzio (2017), without neglecting the role of external actors, take an inside-out approach and frame the regional integration choice of Ukraine with reference to peculiar domestic conditions, such as re-election concerns of incumbent elites, socio-economic structures and their embeddedness in a broader interdependence framework, and national identity.
This article builds on the inside-out perspective concerned with Ukraine, but it opts for regional coverage and embraces a different theoretical approach. Since choices between the EU and EAEU are instances of decision-making under conditions of risk, it imports insights from prospect theory for the study of these integration choices in the EaP region. Importantly, the article does so in the tradition of the philosopher of science, Karl Popper (1963), known in the literature as 'naïve falsificationism', which implies theory testing against empirical evidence. The alternative approach of testing prospect theory against evidence and competing theories approximates what Imre Lakatos (1970) calls 'sophisticated falsificationism' and can be the subsequent superior logical step of theory testing. The current article starts from the premise that the debate between Popper and Lakatos on falsifiability should not be viewed in zero-sum terms, but as distinct, yet legitimate, stages of theory testing. Popperian testing might serve as the first stage of theory falsification, whereas subsequent Lakatosian testing can start from the already known relationship between prospect theory and evidence, and evaluate this relationship in a more complex ontological framework. This article bears the signs of Popperian testing of prospect theory and does not imply that alternative theories, such as (neo)realism and constructivism, are inferior.
In addition, the article employs the ontological repertoire of prospect theory for two idiosyncratic and related reasons. First, it brings into sharper focus the reference point as a crucial concept in prospect theory and, second, it introduces a related multi-attribute reference point (MARP) concept for the study of integration choices in the EaP region. The question this article interrogates is the following: To what extent can the coordinates of MARPs anticipate risk propensities and integration choices of ruling elites from EaP countries?
The location of MARP is examined for all EaP countries and for the years 2013 and 2014. In 2013, the leadership of Armenia and Ukraine suspended permanently or temporarily the association process with the EU and in 2014, Georgia, Moldova and Ukraine signed AA with the EU, while Armenia and Belarus have signed the treaty establishing the EAEU.
Prospect theory has been articulated in the last quarter of the 20th century as a theory of individual decision-making under conditions of risk (Kahneman and Tversky, 1979). Since then, it has become an influential theory in many academic disciplines (Dreher, 2007;Horowitz, 1998;Jolls et al., 1998;Redelmeier and Tversky, 1990;Thaler, 1980Thaler, , 2015. International Relations (IR) scholars have also realised the potential of prospect theory to describe behavioural patterns in international politics and have sought to import prospect-theoretic insights into their subdiscipline (Berejikian, 1997;Farnham, 1994;Jervis, 1994;McDermott, 2001;McInerney, 1992). Applications of prospect theory to IR and foreign policy analysis (FPA) have proved to be beneficial, but they tend largely to reflect the research interests of American scholars and to be limited, hence, to US foreign policy (Mercer, 2005). The way in which ruling elites from EaP countries were forced to opt for distinct integration options offers another possibility for testing prospect-theoretic propositions in an essentially different policy, strategic and regional context.
Despite the analytical centrality of the reference point concept, prospect theory, as originally proposed by Kahneman and Tversky (1979), says little about its identification and tends generally to view it as exogenously given. In addition, the existing applications of prospect theory to IR and FPA often engage this central concept insufficiently and underspecify the dimension(s) on which the reference level may vary (Levy, 1997). This article seeks to address this limitation.
In addition, it introduces the notion of MARPs and provides an example of identifying their coordinates, against which elites from EaP countries are expected to consider alternative regional integration choices. The article proposes to view the coordinates of these reference points as varying on the following three core dimensions: economic affluence, the history of conflict with Russia and the domestic costs of (democratic) transformation. The theoretical prediction informed by prospect theory is to expect risk-seeking propensities in cases where the reference points are coded in the domain of losses. This behavioural inclination appears to be confirmed by Georgia, Moldova and Ukraine in 2014, which have MARP with low coordinates on all three dimensions and where ruling elites proved to be risk-acceptant while opting for integration into the EU.
This article is structured in four sections. The first two sections revisit the concept of reference point, as defined in prospect theory, and review several applications of this concept in political science literature. The third and fourth sections of the article introduce the notion of multi-attribute reference points, identify their coordinates and discuss their implications for risk propensities and integration choices of ruling elites in EaP countries.

Prospect theory and reference point
Prospect theory was articulated in 1979 by psychologists Daniel Kahneman and Amos Tversky as a descriptive theory of individual decision-making under conditions of risk (Kahneman and Tversky, 1979;Tversky and Kahneman, 1992). The theory builds on principles from cognitive psychology, but remarkably it has had the largest impact in economics (Thaler, 1980(Thaler, , 2015. Prospect theory has been elaborated in opposition to expected utility theory (EUT), which was the dominant economic theory of decision-making under risk and is still widely accepted as a normative model of rational choice. Coming from study of heuristics and cognitive biases, Tversky and Kahneman (1974) acknowledge the normative relevance of EUT, but also underscore its limited descriptive capacity. What essentially distinguishes prospect theory from EUT is that the former posits that the main carriers of value are changes in wealth seen as either gains or losses from a neutral reference point. This is the cornerstone insight of prospect theory, compared with EUT, according to which utility is defined in terms of final states of wealth.
Along with the principle of loss aversion, which concerns the asymmetry between losses and gains (Kahneman and Tversky, 1979), another important contribution of prospect theory to decision sciences is the introduction of the reference point concept. The utility function introduced by Bernoulli (1954) was void of the reference point concept, and though it could predict risk aversion, it could not explain risk-seeking behaviour of individuals for gambles with negative prospects (Kahneman, 2011). For Kahneman and Tversky (1979), both coming from the field of psychology, where reference dependence is pervasive in perception and judgement, this theoretical void was obvious. They underline that individuals tend to code outcomes as either gains or losses and to evaluate both positive and negative outcomes as shifts away from a given neutral reference point.
The reference point is called status quo when it corresponds to the current asset position, but it can equally be a past entitlement or future expectation (Kahneman and Tversky, 1979). If an investor is susceptible to the sunk-cost fallacy and continues to operate from a reference point ex-ante, she would be dissatisfied with the current status quo, perceive it as a loss and tend to be risk-seeking.
The reference point might have a relatively static nature, such as the size of one country's territory, as well as a more dynamic one, such as the annual rate of economic growth. The non-static nature of the reference point might also be due to subjective framing effects, which can alter it by alternative formulations of the same decision problem. Because variation of the reference point can turn gains into losses and vice versa, it can change the order of preferences (Tversky and Kahneman, 1992). In opposition with normative theories of decision-making and the axiomatic principle of invariance, prospect theory sees framing effects as being potentially responsible for rationally inconsistent choices (McDermott, 2001).
Before introducing the notion of MARP and the dimensions on which the former can vary, the next section reviews several applications of reference point concept in political science literature.

Reference point concept in political science
There is a largely held consensus that identification of the reference point in applications of prospect theory to political science topics is critical (Jervis, 1994;McDermott, 2001;Mercer, 2005). This agreement notwithstanding, locating the reference point remains a vague enterprise in some of these applications (Levy, 1997). However, this is also the case in other academic disciplines, including Kahneman and Tversky's original body of work.
Political scientists critically observe that prospect theory lacks an elaborate account of how actors locate the reference point (Butler, 2007: 229;Levy, 1997: 100;O'Neill, 2001: 621). Levy underscores in this regard that prospect theory is 'a reference-dependent theory without a theory of the reference point'. Whereas this critique is generally true, it is also unrealistic to expect prospect theory to offer a mechanism for reference point identification which could be used universally across academic disciplines as different as political science and neuroscience. Identification of the reference point should not be viewed as a mechanistic algorithm ready to take input data from any discipline and to return a reference point each time this algorithm concludes. Nor should one expect a universal account for the identification of the reference point in any particular discipline, including political science. Identification of the reference level is a context-dependent process and should be approached on a case-by-case basis.
In his analysis of the European Community's behaviour before and during the Montreal Protocol negotiations to limit the release of ozone-destroying chlorofluorocarbons (CFCs) into the atmosphere, Berejikian (1997) gives the example of a dynamic reference point. It shifted as a result of changing 'several aspects of the political topography' concerning unilateral action by the United States, the emerging scientific consensus and the reversal in the position of US producers on the desirability of a binding protocol. Farnham (1994) gives another example of a dynamic reference point in the context of the Munich Crisis. She argues that '[. . .] Roosevelt's reference point was a European political situation which did not threaten the U.S.'. Whereas, before the Godesberg meeting, Roosevelt believed that even war did not constitute a severe threat to American interests; after Godesberg, his reference point shifted upwards, having reframed the outcome of war in Europe as a catastrophe and viewing it as a departure from the status quo.
McDermott (2001) attempts to define the reference points in the following four peculiar political contexts: the Iranian hostage rescue mission, the decision to admit the Iranian shah on the territory of the United States, the U-2 crisis and the Suez crisis. In all four contexts, the initial reference points have changed. In the first context, the initial reference point of President Carter was the status quo ex-ante when no Americans were held as hostages. This reference point changed instantly when representatives of the American embassy in Tehran were taken captive and President Carter became highly risk-prone, an attitude which resulted in one of the biggest failures in the history of American rescue operations.
In the second context, the change of Carter's position on the opportunity to accept the Iranian shah on American soil resulted again from a shift of the reference point given the sudden drop in domestic and international support. Domestic support fell significantly by October 1979, when the Shah was admitted to the United States. McDermott thus defines the reference point in relation to the 'central decision-maker', which in her first two examples is President Carter.
With reference to the U-2 crisis, McDermott (2001: 109) argues that the reference point shifted as abruptly as in the case of the Iranian hostage rescue mission. Once U-2 was shut down, Eisenhower's reference point plunged instantly into the domain of losses. In the fourth strategic situation, the initial reference point was the status quo ex-ante, when the Suez Canal was under British and French control. Nationalisation of the Suez Canal by Nasser brought Britain and France into the domain of losses, but it did not significantly affect Eisenhower's reference point, which was in a relative domain of gains. McDermott et al. (2002) designed an experiment on weapon procurement and set two different reference points, instructing their subjects to strive for 'superiority' or 'parity' in the weapons balance. They find that striving for superiority tends to increase weapons procurement, as a leader striving for a higher status will see himself in the domain of losses and will be inclined to be risk-seeking and boost defence spending.
McInerney (1992) examines the Soviet policy towards Syria in [1966][1967] and seeks to answer the question of why Soviet leadership chose to risk war in the Middle East. She argues that the Soviets were risk-acceptant in order to secure the existent status quo with the neo-Ba'ath ideologically close pro-Soviet Provisional Command in Syria. This historical example indicates not only the readiness to take on greater than usual risks in order to protect the status quo, but also confirms that adaptation to a superior reference point is quicker than adaptation to an inferior one.
The above examples of reference points in political science literature are dynamic, but they are all framed as uni-dimensional. The next section, drawing on the above pioneering applications of the reference point concept to IR, suggests that the reference points can also be conceived as multi-dimensional.

MARPs and hypothesis
Unlike the original version of prospect theory where the reference point is usually unidimensional, in non-experimental complex social settings, such as international politics, the reference level can also be multi-dimensional. Cross-cutting dimensions have most likely established the reference positions of alliance-seeking belligerents in times of war, competing factions in revolutions, parties to the Non-Aligned Movement and contenders in ideological battles. Similarly, country-specific MARP are expected to influence risk propensities and distinct integration choices of ruling elites from EaP countries.
Importantly, risk is defined here relative to variance in the outcome (McDermott, 2001: 39-40;Vis and Kuijpers, 2018: 577). The higher the disparity between the expected gain and potential loss, the riskier the option implying this disparity. In the light of this qualification and with the advantage of hindsight, the EU as an integration option is riskier than the alternative EAEU. Economists concur that in the long run economic gains would be higher with the EU as an integration choice (Adarov and Havlik, 2016). At the same time, it is already clear that the choice for the EU might also incur greater losses, especially in a short-and medium-term perspective. Ukraine lost Crimea, control over Donetsk and Luhansk oblasts and several thousands of servicemen and innocent civilians. On the contrary, integrating with the EAEU proved to be less risky. In 2013, Armenian leadership suspended association with the EU and opted instead for integration into the EAEU. However, instead of adopting a coercive approach, the EU has surprisingly provided the highest volume of financial assistance to Armenia in 2015, the year when the latter formally joined the competing EAEU.
The reference position, against which ruling elites from EaP countries are expected to select distinct integration options, can be conceived of as varying on the following three core dimensions: economic affluence, the history of conflict with Russia and the domestic costs of (democratic) transformation. Economic affluence and the history of conflict with Russia correspond to the main economic-and security-based variables in liberal and realist theoretical traditions. For liberals, domestic actors and groups promote transnational and intergovernmental cooperation to advance their own individual prosperity (Moravcsik, 1998;Russett and Oneal, 2001). Realists argue that states do not cooperate, but compete with one another for security and with a view to exert or resist coercion (Waltz, 1993(Waltz, , 2000. The dimension of domestic costs of transformation draws on the external incentives model (EIM) (Schimmelfennig and Sedelmeier, 2004). The costs of rule adoption in the original EIM, along with the credibility and size of EU's external incentives, account for variation in democratic rule transfer to pre-accession Central and East European Countries (Schimmelfennig, 2005). Suggested by distinct theoretical traditions, the above mentioned three dimensions are used to approximate and locate the coordinates of the multi-attribute reference points, against which incumbent elites from EaP countries are expected to manifest different risk attitudes. Adapted to multiple attributes, the prospect-theoretic informed hypothesis this article seeks to test against empirical evidence can be formulated as follows: The lower the coordinates of the MARP as perceived by national governing elites, the more likely would be their risk-seeking propensity and the choice of the EU as an integration project.
Interviews with elites and policy experts 1 and secondary sources suggest that the salience attached to economic affluence, the history of conflict with Russia, and the domestic costs of transformation may vary: across countries, across elites and public within countries and across groups within elites and public. This variation of salience is accounted for with corresponding qualifying observations in the subsequent 'Discussion' section of the article.

Economic affluence
Liberal theories argue that states engage in international cooperation to advance their primary economic goals. These theories put in the centre of their analytic introspection the role of domestic politics and groups, which are interested in their individual economic prosperity. The intergovernmentalist school of liberal tradition starts from the central role of domestic context and actors, the latter being viewed as acting simultaneously on two levels: domestic and supranational. Moravcsik (1998) argues in this regard that it remains at the discretion of national governments to decide which issues to be negotiated at supranational level and that they do so in the interest of their domestic constituencies. Liberal intergovernmentalist accounts suggest that regional economic integration tends to remain a process under strict governmental control and advances only when this process satisfies national economic preferences.
In a broad historical perspective, pursuing economic prosperity has been one of the most stable trends in human history. McDermott et al. (2008) confirm that pursuing economic affluence has a prominent evolutionary origin. Indeed, the main assets in primitive tribal societies were the tools used by hunter-gatherers for their rudimentary activities. Following the agricultural revolution, land became the main economic asset, whereas the means of production were the main assets in the industrial era. In today's post-industrial societies, means of production are being overtaken by knowledge and data as the main economic asset (The Economist, 2017). Tools, land, means of production and data vary significantly in their material versus non-material and static versus dynamic nature, but they are always used to the same end: ensuring increased economic prosperity.
Regional integration processes, which currently engulf EaP countries, can also be seen as part of the same general trend of pursuing economic modernisation and prosperity (Dragneva and Wolczuk, 2015). All countries in the region, except Azerbaijan, have generally scarce domestic resources and elites from these countries have tended to view participation in regional integration projects as a complimentary way of ensuring economic prosperity. At the beginning of the 1990s, the economic contraction in these countries was so severe that entire industries either vanished or are still struggling to recover. The gross domestic product (GDP) fell by 42% in 1992 in Armenia, 45% in 1992in Georgia, 31% in 1994in Moldova and 23% in 1994in Ukraine (World Bank, 2018. The economies of Belarus and Azerbaijan also contracted significantly in the same period, but they reversed the trend with robust economic growth. For three consecutive years in 2005-2007, the Azeri economy was the fastest-growing economy in the world, with growth rates of 26.4%, 34.5% and 25.1% (World Bank, 2018). The extraordinary rise of GDP per capita from US$2420 in 1995 to US$17,257 in 2016 and the high average GDP growth rate in Azerbaijan became possible with a more intense exploitation of oil reserves. The resulting relative affluence allowed Azerbaijan to depend less on external resources and to choose the terms on which to interact with both the EU and Russia. The average annual volume of external assistance provided by international donors to Azerbaijan over the period 1991-2016 is comparable with the amount Azerbaijan could earn in only a few days from oil revenues (Buscaneanu, 2016). Given the availability of internal resources, Azerbaijan has less stringent need to search for external resources.
Belarus could count less on internal natural resources, but its leadership under Lukashenko was relatively successful in maintaining and restructuring several industries, such as heavy machinery, which were important suppliers for the centrally planned Soviet economy (Charap and Colton, 2017). At the time of the dissolution of the USSR, Belarus was one of the most industrialised and most affluent Soviet republics, a potential which has been maintained up to the present day. Belarus has a slightly higher GDP per capita than Azerbaijan (World Bank, 2018), but in contrast to the latter, it depends heavily on Russia as its main trade market and as the main supplier of cheap energy resources for its energy-intensive industrial sector. In fact, one of the main reasons why sectors such as heavy machinery proved relatively successful in Belarus was the possibility to import subsidised energy resources from Russia, keeping the production costs low and exporting finite industrial products to Russian and Commonwealth of Independent States (CIS) markets at competitive prices (Charap and Colton, 2017).
Moldova, Armenia, Georgia and Ukraine were less successful than Belarus in readjusting their economies in the new post-Soviet context and have fewer internal resources than Azerbaijan. In a prospect-theoretic application, McDermott et al. (2008) demonstrate that the striving for economic prosperity has implications for risk attitudes, with riskless choices in economically rich environments and risky ones in poor environments. Similarly, being less affluent than Belarus and Azerbaijan and one of the least affluent countries in Europe is expected to pull down the corresponding reference points of Moldova, Armenia, Georgia and Ukraine in the domain of loss and to make their striving to economic affluence particularly stringent.

Conflict with Russia
Realists tend to view international politics as an interstate struggle for security. For them, military power is the most compelling form of power in the anarchic international system, which states pursue in order to coerce or resist being coerced by other states (Grieco, 1988). Prospect-theoretic applications in IR borrow insights from realist accounts and confirm that conflicts have profound implications for the risk attitudes of involved parties (Butler, 2007;Farnham, 1994;Jervis, 1994;McDermott, 2001).
As the future is viewed through the prism of the past, future friendships are viewed through the prism of past conflicts (Mercer, 2005). Similarly, the intensity of conflict with Russia can be viewed as affecting the readiness of target states to take on risks and might also serve as a crucial predictor of the present and future relationships between EaP countries and Moscow.
Georgia, Moldova and, recently, Ukraine have experienced military conflicts with Russia and have secessionist regimes backed by the Kremlin on their territories. The perception of traumatic military conflict with Russia is particularly acute in Ukraine. The Russian annexation of Crimea, with an area similar to that of Armenia and a population similar to that of Latvia, was the first territorial acquisition by brute military force since the Second World War (Plokhy, 2017). Recall in what follows one of the most blatant strategic miscalculations in recent history. The Ukrainian President Yanukovych agreed in 2010 to prolong the Russian Black Sea fleet's lease of the Sevastopol naval base until 2042 in exchange for a gas price discount (Dragneva and Wolczuk, 2015;Wilson, 2014). The extension was also meant to reverse the consequence of Tymoshenko's 'abuse of office', sentenced to 7 years in prison for, allegedly, concluding an unfavourable deal on the price of gas with Gazprom in 2009. In 2014, Russia made use of the Black Sea naval base to annex Crimea, and in the new post-Maidan context the price discount agreement with Gazprom became obsolete. Ukraine lost a territory of 27,000 km 2 , the agreed gas price discount and the substantial fee the Russian government was expected to payapproximately US$90 million annually -for lease of the base.
Large segments of the general public and elites from Georgia, Moldova and Ukraine perceive Russia as an aggressor and as 'the Damocles' sword' hanging menacingly over the security of their statehood. In these countries, being pro-Russian is often stigmatised and a pro-EU or, more generally, pro-Western attitude is a political and cultural statement that often substitutes the ideological quintessence of a political party's manifesto (Charap and Colton, 2017). The single most important criterion on which political parties in Georgia, Moldova and Ukraine differ is not their ideological stance or position on taxes, wages, spiritual values and LGBT rights, but their attitude towards the West and Russia. It is the latter attitude which often signals the former. Geo-political rather than political is therefore the more appropriate adjective describing Georgian, Moldovan and Ukrainian parties.
Armenia and Belarus have had no military conflicts with Russia, nor do they have secessionist regimes on their territory backed by the Kremlin. In addition, since 1994, Armenia and Belarus have been members of the Russian-led Collective Security Treaty Organisation (CSTO), a military alliance conceived to offer mutual military support in case of external aggression. Russian security backing is particularly valuable for Armenia, which has a long-term dispute with Azerbaijan over Nagorno-Karabakh and seeks Moscow's support to counterbalance the strategic weight of Baku (Charap and Colton, 2017).
The dispute between Azerbaijan and Armenia over Nagorno-Karabakh explains partly why Azeri-Russian relationships are not free of difficulties. In contrast to Georgia, Moldova and Ukraine, Azerbaijan does not have a direct conflict with Russia. However, given the fact that important segments of Azeri public and elites perceive the Kremlin as being sympathetic to Yerevan's position on the status of Nagorno-Karabakh, they tend to regard Moscow with circumspection (Abushov, 2019). One consequence of this circumspection is the distance Azeri elites choose to take from Moscow and the highly selective participation of Azerbaijan in CIS initiatives. Additional empirical evidence supports this reluctant attitude. Azerbaijan has the fourth lowest rate of ratifications (−0.44 standard deviation (SD) below the mean) and the second lowest rate of signatures (−0.85 SD below the mean) of multilateral agreements among all CIS countries in the period 1991-2020.
Ukraine, Georgia and Moldova have a history of traumatic conflict with Russia, which is expected to pull down their corresponding reference points into the domain of loss. The perception of the Azeri public and elite with regard to the indirect low conflict with Russia is expected to have a slight downward effect on Azerbaijan's reference level. In contrast, Armenia and Belarus have no major conflicts with Russia and are expected to have their corresponding reference points on this dimension in the domain of gain.

Domestic costs of transformation
Domestic costs feature as a core variable in the EIM of governance, which was developed by Schimmelfennig and Sedelmeier and explains variation in democratic rule transfer to Central and East European Countries (Schimmelfennig, 2005;Schimmelfennig and Sedelmeier, 2004). Schimmelfennig and Sedelmeier (2004) evaluate these costs depending on the ideological outlook of party constellations. This article builds on the EIM, but approximates domestic costs of transformation with the theoretical distance from the empirical state of a given political regime to the theoretical value at which political regimes can be considered democratic (see Appendix A1).
EaP countries populate the spectrum covered by autocratic and hybrid regimes (Hale, 2015;Way, 2015). Azerbaijan and Belarus have entrenched autocratic regimes, whereas regimes in Armenia, Georgia, Moldova and Ukraine exhibit both democratic and autocratic qualities. Regime types have direct implications for the size of domestic costs of transformation, which tend on average to be higher in the case of autocratic regimes than in hybrid ones. Azerbaijan and Belarus are expected thus to face higher costs of transformation than Armenia, Georgia, Moldova and Ukraine, where these costs, though lower, would still represent non-trivial impediments to democratic consolidation. The non-trivial nature of impediments to full-fledged democratisation in competitive authoritarian regimes of Armenia, Georgia, Moldova and Ukraine is conclusively documented by Way (2015) and Hale (2015).
EaP countries are dominated by political regimes where informal institutions, such as patronage, nepotism and clientelism, have a central role in the access to and the exercise of power (Hale, 2015). Informal institutions have a long evolutionary history, are highly resilient to change, and are natural forms of social organisation in EaP countries. They are not the consequence of poor governance, but rather the symptom of political underdevelopment (Fukuyama, 2015).
Patronage is about single or multiple pyramids of authority with patrons at the top and their clients at different layers of the administrative system (Hale, 2015). The latter consent to offer loyalty for and subordination to the patron, in exchange for rents and favours they can extract while using their administrative positions in ministries, agencies, courts, law enforcement institutions and state-owned enterprises. The higher the rents extracted by the elites, the higher their material gains. The higher the material gains of the elites, the higher the costs they anticipate with genuine political liberalisation and democratic consolidation. The higher the potential costs faced by these elites, the higher their resistance to any reform process aiming at democratisation is.
The domestic costs of transformation are therefore those costs which threaten directly the hold on power chances of (semi)autocratic elites. Such costs threatening the hold on power are expected with regime liberalisation and implied directly by the EU's democratic conditionality. These costs should not be confused with regulatory costs, which are implied by the EU's acquis conditionality and expected with regulatory approximation and convergence (Schimmelfennig and Sedelmeier, 2004). It is the former type of costs that render the democratisation process in EaP countries difficult.
Given the fact that the EU and EAEU have very different attitudes towards democracy and autocracy, with the former conditioning European integration with genuine democratic reform, the costs of transformation are expected to be higher with the EU as an integration choice and especially prohibitive for elites from Azerbaijan and Belarus, where autocratic practices tend to be stronger. To be sure, Armenia, Georgia, Moldova and Ukraine also face non-trivial costs of transformation, but in contrast to Belarus and Azerbaijan, they are expected to have lower reference levels on this dimension.
Economic affluence, the absence or presence of conflict with Russia, and variable costs of democratic transformation jointly signpost the location of MARP, which are expected to anticipate distinct risk propensities and regional integration choices of incumbent elites from EaP countries (see Figure 1).

Discussion
The coordinates of the above MARP in EaP countries, against which distinct risk attitudes and regional integration choices are expected, are context-dependent. The choices of ruling elites from EaP countries to embrace distinct regional integration projects took place in a dynamic material and security context, which defines the coordinates of cross-country MARP against which alternative regional integration outcomes could be considered.
Except for Ukraine, the reference points of EaP countries have approximately the same location in 2013 and 2014. Although Belarus and Azerbaijan are more affluent than the rest of the EaP countries, in 2013 and 2014 they remain below the average of Organisation for Economic Co-operation and Development (OECD) countries in terms of per capita wealth and are also expected to look out for external economic resources. This inclination is expected, however, to be more prominent in the case of Belarus, which is highly dependent on the Russian trade market and economic subsidies (Charap and Colton, 2017;Dragneva and Wolczuk, 2015). On the same dimension of economic affluence, the reference points of Moldova, Armenia, Georgia and Ukraine are in 2013-2014 even lower than those of Belarus and Azerbaijan. Hence, their search for external economic resources, including through participation in regional integration initiatives, is expected to be even more pronounced.
On the dimension of conflict with Russia, in 2013, Georgia and Moldova were in a state of medium intensity dispute. Tbilisi and Chișinău did not experience military confrontation with Moscow in 2013, but continued to have separatist regimes backed openly by the Kremlin on their territories. Unlike Georgia and Moldova, Azerbaijan has no direct conflict with Russia, but given the perceived Russian support to Armenia on the status of Nagorno-Karabakh (Abushov, 2019), its corresponding reference point in 2013 is located below the reference points of Armenia, Belarus and Ukraine, which had no conflicts with Russia. In 2014, the reference point of Ukraine shifted abruptly from the front lower right to the front lower left section of the property space (see Figure 1), given the sudden loss of Crimea and the violent military conflict with Russia in the Donetsk and Luhansk oblasts. Ukraine is the only case whose reference point changed so abruptly from 2013 to 2014 on the dimension of conflict with Russia. Ruling elites from Georgia, Moldova and, especially, those from Ukraine are then expected to be risk-seeking, whereas those in Armenia and Belarus are expected to be risk-averse vis-à-vis Russia. The risk propensity of Azerbaijan on this dimension is ambiguous and expected to depend on the readiness of Kremlin to accommodate Baku's strategic interests (Abushov, 2019).
On the domestic costs of transformation, incumbent elites from Belarus and Azerbaijan would face the highest costs, whereas Armenian elites would face medium costs in the event of transforming their regime along a democratic path, which is the main prerequisite for joining the EU. On this dimension, Georgian, Moldovan and Ukrainian ruling elites would face lower costs of democratic transformation, which however remain nontrivial. Here again, Ukraine is the only nation among all EaP countries whose costs of transformation shifted downwards in 2014 compared with 2013. Resistance to democratic reforms is expected be the highest in Belarus and Azerbaijan, medium in Armenia and lower, but still not trivial in the rest of the EaP cases.
Tri-dimensional representation of the reference points of EaP countries illustrates in line with the hypothesis that cases with the lowest level of affluence, with medium to high intensity conflicts with Russia, and with lower costs of domestic transformation have tended to be risk-seeking and embrace the option of European integration. These cases are Georgia, Moldova and Ukraine. The latter forcefully joined this group of countries in 2014, when its corresponding reference point shifted severely on the dimension of conflict with Russia and moderately on the costs of transformation. Although conflicts maintained by Russia in Georgia, Moldova and Ukraine might serve as a strategic bargaining chip for the Kremlin, they appear to have self-defeating consequences in the long run: the more intense the conflict is, the more targeted countries tend to drift away from the Russian-led EAEU. Russia acquired Crimea, but it might have lost for an indefinite time the friendship and loyalty of Ukraine (Kuzio, 2017;Plokhy, 2017). Similarly, the Kremlin exerts tight control over the separatist entities in South Ossetia, Abkhazia and Transnistria, but it is not credited with much credibility and trust in Georgia and Moldova (Buscaneanu, 2016). However, cases with low and medium levels of affluence, but importantly, with no experience of violent military conflict with and no secessionist regimes on their territory backed by Russia, and with medium to high costs of domestic transformation have tended to be risk-averse and embrace the EAEU as an option for regional integration. These cases are Armenia and Belarus.
Finally, the case with a medium level of economic affluence, a low level of conflict with Russia, though still perceived by the elite and general public, and with high costs of domestic transformation has tended be risk-neutral and to stay away from both regional integration projects. This case is Azerbaijan. In the light of hypothesis and empirical evidence, Azerbaijan is expected to display risk-averse behaviour, since the location of its reference point is fairly similar to that of Belarus on the dimensions of economic affluence and costs of transformation. However, it differentiates moderately on the dimension of conflict with Russia, which can potentially explain the risk-neutral inclination of Azeri ruling elites. A solution to the Nagorno-Karabakh conflict, which would appease all of Baku's chief grievances, could shift Azerbaijan's reference point upwards and make Baku more ready to consider an integration offer from the EAEU (Abushov, 2019). However, Azeri ruling elites are not as risk-seeking as those from Georgia, Moldova and Ukraine, because Baku is less dependent on EU's financial assistance, and association with the EU -conditional on political liberalisation -would impose high costs threatening the elites' grip on power.
Empirical evidence from interviews and secondary sources suggests that the salience attached to the dimensions capturing the conceptual role of MARP may vary across countries. For instance, in the light of hypothesis, Armenia is expected to manifest an ambiguous risk propensity, as it is less affluent and would face lower costs of transformation than Belarus. However, among all EaP countries, Armenia attaches the greatest salience to the security-related dimension. The acutely felt security problem over Nagorno-Karabakh and the strategic and military backing expected from Russia, including through CSTO, are potentially sufficient to account alone for the risk-averse inclination and the integration choice of Armenian ruling elites. 3 Any medium to high intensity conflict with Moscow would potentially shift the reference point corresponding to Yerevan in the same section of the property space where the reference points of Georgia, Moldova and Ukraine lay in 2014 and alter its risk propensity, respectively. Furthermore, evidence from Ukraine suggests that the salience attached to economic affluence and costs of transformation can also vary across elites and public, and across distinct groups within elites and public. Ukrainian interviewees indicate unanimously that a risk-averse behaviour on the part of those elites supporting Yanukovych was expectable, as the lion's share of their wealth originated from opaque transactions on the Russian market and since they would also be expected to face higher costs in the event of political liberalisation, implied by association with the EU. However, this risk propensity was not shared by an opposing segment of the elite, connected less with Russian business interests, and by a significant part of the Ukrainian public in dire need of economic and political modernisation. 4 Two additional observations on regional integration U-turns in the EaP region are in order. First, Georgia and Moldova had the lowest reference points in 2013 and they are the cases, which, unlike Armenia and Ukraine, did not experience integration U-turns in the face of Russian economic and the strategic pressure meant to derail their association with the EU. In line with the prediction of prospect theory, Georgian and Moldovan elites proved to be risk-acceptant and went on with initialling and then signing the AA with the EU despite significant Russian pressure in 2013 and 2014.
Second, the reference positions corresponding to Armenia and Ukraine in 2013 are similar to those of Georgia and Moldova on the dimensions of economic affluence and costs of transformation. However, where the former cases differ significantly is the intensity of conflict with Russia. Having no major conflicts with Moscow and their reference points in the domain of gain on this dimension before 2013, the expected propensity informed by prospect theory would be risk aversion. Indeed, Armenia and Ukraine proved to be risk-averse, and conceded in the face of Russian pressure, suspending completely or temporarily the association process with the EU.
One influential account suggests that Yanukovych's immediate concerns for his reelection with Russian support in the approaching 2015 election campaign did also play a role in the decision of suspending the association process with the EU (Wilson, 2014: 80). Whereas, this rationale remains difficult to be objectively tested, it should be noted that Armenian President Sargsyan and his entourage took a similar decision under more relaxed time constraints until the next nation-wide elections. At the time of regional integration U-turn in 2013, Armenian incumbent elites were facing a long time horizon until the next presidential (4.42 years) and legislative elections (3.58 years), whereas the pressure of re-election concerns should have been ambiguous for Ukrainian ruling elites confronted with both short-and long-termness before the next presidential (1.2 years) and legislative elections (3.86 years). In short, Armenia and Ukraine experienced similar U-turns under different time horizons until the next nation-wide elections. However, what both Armenia and Ukraine had in common before 2013 was the absence of severe territorial conflicts with and losses inflicted by Russia.
After severe loss of territory, secession in the Donetsk and Luhansk oblasts, the ongoing covert war with Russia and several thousands of human losses, the reference point corresponding to Ukraine shifted abruptly in the domain of net loss. Under new circumstances of severe losses, the expected inclination informed by prospect theory would be risk-seeking. For the new post-Maidan leadership under Poroshenko, fully aware of further economic sanctions and costly military pressure from the side of Moscow, the decision to bring back on the table and to sign the AA with the EU was out of the question. The same risk-acceptant inclination can be witnessed during Zelensky's presidential mandate and is expected to factor in the decision-making of Ukrainian incumbent elites for many years to come, unless Kremlin decides to change course and redress the unfortunate history of interstate conflict.

Conclusion
'Of what you have found you do not speak; / (Only) of what you have lost do you speak' (Gordon, 1959: 47). This reproach to pessimists sounds fairly familiar to anyone living in 2020 in the EU, EaP countries or Russia. In fact, the above reproach is a Sumerian proverb and was highly familiar to Sumerians living some 5000 years ago. Five thousand years of human evolution has left the human inclination to discount gains and overweigh losses completely unchanged. In 1979, Daniel Kahneman and Amos Tversky made sense of and defined this inclination as loss aversion.
Apart from loss aversion, the second major contribution of the joint work of Kahneman and Tversky to the study of individual decision-making under risk was the introduction of the reference point concept. First imported into the field of economic theory by Richard Thaler (1980), the reference point concept has also crossed into the disciplinary boundary of political science. However, the reference point concept remains outside the primary focus of prospect theory and is often underspecified in existing applications of prospect theory in IR and FPA. This article makes a special case for the application of prospect theory's reference point concept to a new policy, strategic and regional context. The article examines the extent to which the reference point concept can anticipate risk propensities and integration choices of EaP countries from the vantage point of the Popperian tradition of theory falsifiability against empirical evidence.
The article introduces the notion of MARPs and identifies their coordinates, against which ruling elites from the EaP region are expected to manifest distinct risk propensities and to consider distinct integration choices: the EU or EAEU. This article suggests that the coordinates of country-specific reference points can be viewed as varying on three core dimensions: economic affluence, the history of conflict with Russia and the domestic costs of (democratic) transformation.
Identifying the location of contextual MARP appears to be a compelling analytical tool, which allows risk inclinations and integration choices of ruling elites from the EaP region to be anticipated. Similarly, students of policy choices in comparative politics and international outcomes in IR might use the concept of MARP to bring risk propensities into sharper focus. Compared with uni-dimensional reference positions usually applied in the ontological realm of behavioural economics, MARP might better serve the study of phenomena in political science, where the presence of cross-cutting dimensions often defines the choice made between competing sides in wars, revolutions and ideological struggles.
This article finds in line with the prospect-theoretic informed hypothesis that ruling elites from EaP countries with the lowest levels of economic affluence, with medium to high intensity conflicts with Russia and with lower, but still non-trivial costs of domestic transformation have tended to be risk-seeking and opted for the EU as an integration choice. These countries are Georgia, Moldova and Ukraine. However, incumbent elites from EaP countries with low and medium levels of affluence, but, importantly, with no conflicts with Russia and with medium to high costs of domestic transformation have tended to be risk-averse and selected the EAEU as an integration option. These cases are Armenia and Belarus. Finally, ruling elites from Azerbaijan with a medium level of affluence, low intensity conflict with Russia and high costs of domestic transformation have an ambiguous risk attitude and have tended to stay away from both regional integration projects. Tertium datur.
What is the added value of prospect-theoretic insights for the study of regional integration choices? The conceptual apparatus of prospect theory opens a new analytical angle from which the process of regional integration can be approached. Beyond cost-benefit calculations and identity-based considerations, advanced by realist, liberal and constructivist accounts, prospect theory emphasises the analytical role of contextual reference points, loss aversion and risk propensities factoring in the binary selection of regional integration projects. Prospect theory thereby instructs the students of regional integration to pay attention to states' variable reference positions in multi-attribute property spaces and to their idiosyncratic risk evaluations.
One of the core insights of this article is that country-specific MARP in the EaP region are not indefinitely fixed, but rather have a context-dependent dynamic quality. Shifts of the reference level along the dimensions of economic affluence, conflict with Russia and domestic costs of transformation may bring about corresponding changes in risk attitudes and evaluations of regional integration offers. The latter shall remain firmly on the research agenda along with the superior ontological step of 'sophisticated falsificationism', which implies prospect-theoretic testing against the evidence of regional integration choices and the explanatory power of competing theories.